Guided reflection in collaborative problem-solving: Leveraging natural language processing and learning analytics

This study examines the impact of guided reflection on students’ learning performance, reflective thinking, and collaborative interactions in STEM problem-solving. A reflection-support system integrating natural language processing (NLP) was developed to capture and analyze students’ discussion content in real time. A randomized controlled trial with 76 undergraduates in a Micro:bit autonomous vehicle workshop compared guided reflection prompts with open-ended reflection. Results showed that guided reflection significantly enhanced learning performance and reflective thinking. Semantic analysis and visualizations further revealed that the experimental group adopted more coherent problem-solving strategies and balanced participation. These findings highlight the pedagogical value of guided reflection in collaborative STEM learning and demonstrate the potential of data-informed tools to support instructional decision-making and engagement.